RankingEffort benchmarks

Grok 4.6 · Medium effort benchmarks

Compare measured Low, Medium, High and other settings. Every number keeps its source.

Model

Grok 4.6

98 measured results · 4 reported effort labels · 39 benchmark/harness combinations. Model evidence and sources →

Capability score evidence

Grok 4.6 · Medium: 59.6 overall comparison score. Meets the overall evidence requirements.

Which sources carry the weight?

ARC Prize · 28.9% direct comparison weight · independent
Cognition · 28.9% direct comparison weight · independent
Artificial Analysis · 25.0% direct comparison weight · independent
Datacurve · 17.3% direct comparison weight · independent

These shares describe the comparisons entering the fit, not fractions of the final score.

How sensitive is this score?

Remove one evaluator or family and refit the graph against the same overall panel. Values use the matching dataset audited 2026-09-12. These are sensitivity checks, not confidence intervals; ranking eligibility may also change.

Removed evidence	Recomputed score	Change
Artificial Analysis	59.7	+0.1
ARC Prize	69.6	+10.0
Datacurve	53.4	-6.2
LiveCodeBench	59.6	+0.0
Scale	60.6	+1.0
Terminal-Bench	59.8	+0.2
Cognition	53.8	-5.8
LiveBench	60.0	+0.4
SWE-rebench	59.3	-0.3
Epoch AI	60.3	+0.7
Vals AI	59.1	-0.5
Provider reports · Anthropic	59.6	+0.0
Provider reports · Moonshot AI	59.5	-0.1
Provider reports · Meta	59.6	+0.0
Provider reports · Mistral	59.6	+0.0
All provider reports	59.5	-0.1

Every benchmark family

Removed evidence	Recomputed score	Change
arc-agi	70.5	+10.9
deepswe	53.4	-6.2
frontiercode	53.8	-5.8
mmmu-pro	60.4	+0.8
hle	60.3	+0.7
simpleqa	60.2	+0.6
arc-agi-3	59.1	-0.5
epoch-game-puzzles	59.1	-0.5
automationbench	59.2	-0.4
omniscience	59.9	+0.3
mmlu-pro	59.9	+0.3
terminal-bench	59.9	+0.3
ifbench	59.3	-0.3
swe-rebench	59.3	-0.3
harvey	59.3	-0.3
swe-atlas-qna	59.4	-0.3
multi-swe-bench	59.4	-0.2
livebench-instructions	59.8	+0.2
briefcase	59.4	-0.2
gdp-pdf	59.8	+0.2
tau-bench	59.4	-0.2
aime	59.4	-0.2
livebench-language	59.8	+0.1
swe-atlas-test-writing	59.5	-0.1
critpt	59.5	-0.1
livebench-coding	59.5	-0.1
gpqa	59.5	-0.1
livebench-reasoning	59.5	-0.1
vibe-code	59.5	-0.1
apex-agents	59.7	+0.1
vals-legal-research	59.5	-0.1
aa-lcr	59.7	+0.1
scicode	59.7	+0.1
frontiermath	59.7	+0.1
livebench-data	59.7	+0.1
gdpval	59.6	-0.1
enigma-eval	59.6	-0.0
analyst-agent	59.6	+0.0
vals-code-migration	59.6	-0.0
swe-atlas-refactoring	59.6	-0.0
livebench-math	59.6	+0.0
vals-excel-modeling	59.6	+0.0
livecodebench	59.6	-0.0
vals-finance-agent	59.6	-0.0
gmmlu	59.6	-0.0
math500	59.6	-0.0
itbench	59.6	-0.0
enterprise-ops	59.6	+0.0
swe-bench-pro	59.6	+0.0

What evidence is missing from the fit?

Across all collected settings for this model: 119 observations, 90 contributing, 0 matched without graph weight and 29 excluded. Counts do not establish rank eligibility.

4 · Benchmark family or source protocol has not been reviewed
4 · Evaluator has not been reviewed for independent admission
11 · Supporting evidence outside the reviewed capability core
9 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately
1 · No matched opponent in this evaluation unit

Inspect every observation and exclusion →

Capability profile

One configuration, seven capabilities

Comparison score · 0–100

AGrok 4.6 · Medium3 of 7 capabilities supported

Gaps are unknown, not zero. Hollow points are preliminary.

Agentic
A67.2Supported
Hard reasoning
A47.6Supported
Coding
A71.2Supported
Human pref
Unknown
Knowledge
A35.8Preliminary
Multimodal
Unknown
Long context
A59.6Preliminary

Each profile belongs to the named configuration. Capability scores use a shared panel for all axes, distinct from the overall ranking panel; they are not accuracy percentages or direct head-to-head wins.

Compare this configuration →Explore the model evidence profile →

These are published benchmark scores. Supported effort estimates also appear on the Capability leaderboard. Results from different harnesses occupy different rows. Missing cells stay unknown; no scores are copied between effort levels.

Benchmark / source	Low	Medium	High	XHigh
AA-AnalystAgentArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of tasks solved on all 5 of 5 attempts · 80 tasks, 5 attempts per task · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	41.25% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	Not reported
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Briefcase is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-09-12.	1,293.13 Elo Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	1,490.6 Elo Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	1,534.13 Elo Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	1,545.11 Elo Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
AA-LCR v1.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	80.67% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	81% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	80.33% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	81% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
AutomationBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of task objectives completed with no guardrail violations · Higher is better · Benchmark developed by Zapier · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	53.45% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	63.15% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	66.68% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	67% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
CritPtArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	5.71% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	17.71% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	17.14% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	19.71% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
EnterpriseOps-Gym-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Overall task success rate across all eight enterprise domains (oracle tool mode) · Benchmark developed by ServiceNow Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	48.34% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	Not reported
GDP.pdf: All-passArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of attempts where every atomic criterion passed · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	15.8% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	17.8% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	17% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	17.2% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
GDPval-AA v2 LeaderboardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Elo rating for performance on real-world work tasks · Anchored to a human baseline of 1,000 · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-09-12.	1,461.3 Elo Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	1,642.56 Elo Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	1,642.96 Elo Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	1,663.07 Elo Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
GPQA DiamondArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	87.88% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	93.54% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	94.95% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	93.54% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
Humanity's Last ExamArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	27.62% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	42.12% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	42.91% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	44.07% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of tasks judged accurate, complete, and concise · Expert + compound (hard) sets · Benchmark developed by Wisedocs · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	12.22% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	Not reported
AA-Omniscience IndexArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis Unit: index. Source collected 2026-09-12.	25.9 Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	28 Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	30.48 Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	29.32 Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
AA-Omniscience AccuracyArtificial Analysis · 6000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader Benchmark details AA-Omniscience Accuracy (higher is better) measures the proportion of correctly answered questions out of all questions, regardless of whether the model chooses to answer · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	43.28% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	41.93% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	48.23% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	43% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
SciCodeArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	49.42% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	55.9% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	56.48% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	53.01% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
𝜏³-BankingArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Sierra Research · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	38.14% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	44.33% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	50.72% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	43.3% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
Terminal-Bench v2.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by the Laude Institute and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	75.28% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	84.27% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	88.39% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	88.01% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
Terminal-Bench v4.0Artificial Analysis · mini-SWE-agent v2.4.6; 66 tasks; pass@1 averaged over 3 repeats; 500 steps; task timeout up to 8 hours Benchmark details Benchmark developed by the Laude Institute, Stanford, and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-09-12.	3.03% Reported settings Grok 4.6 (low); Model source: https://artificialanalysis.ai/models/grok-4-6-low	13.13% Reported settings Grok 4.6 (medium); Model source: https://artificialanalysis.ai/models/grok-4-6-medium	21.21% Reported settings Grok 4.6 (high); Model source: https://artificialanalysis.ai/models/grok-4-6	17.17% Reported settings Grok 4.6 (xhigh); Model source: https://artificialanalysis.ai/models/grok-4-6-xhigh
ARC-AGI-2ARC Prize · v2 Semi-Private Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-09-12.	27.64% Reported settings Grok 4.6 (Low); Source model ID: xai-grok-4-6-low; 120 tasks	61.25% Reported settings Grok 4.6 (Medium); Source model ID: xai-grok-4-6-medium; 120 tasks	65.14% Reported settings Grok 4.6 (High); Source model ID: xai-grok-4-6-high; 120 tasks	67.08% Reported settings Grok 4.6 (XHigh); Source model ID: xai-grok-4-6-xhigh; 120 tasks
DeepSWE v1.1Datacurve · mini-swe-agent Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-09-12.	41.6% Reported settings grok-4-6; Reported effort: low; 113 unique tasks; 449 attempts; 4 runs	67.5% Reported settings grok-4-6; Reported effort: medium; 113 unique tasks; 452 attempts; 4 runs	65.2% Reported settings grok-4-6; Reported effort: high; 113 unique tasks; 451 attempts; 4 runs	66.7% Reported settings grok-4-6; Reported effort: xhigh; 113 unique tasks; 451 attempts; 4 runs
VulcanBench v3 · Report 14VulcanBench · VulcanBench bare-bones API loop, deterministic hidden tests in Docker Benchmark details Frozen v3, 23 post-cutoff repository tasks. Published pass@1; repeat counts and time budgets vary by report. Display only pending comparability review. Unit: percent. Source collected 2026-09-12.	82.6% Reported settings Grok 4.6; source label: low; low effort; 23 tasks; evaluation 2026-08-12. See source for repeat counts and wall-clock budget.	87% Reported settings Grok 4.6; source label: medium; medium effort; 23 tasks; evaluation 2026-08-12. See source for repeat counts and wall-clock budget.	73.9% Reported settings Grok 4.6; source label: high (default); high effort; 23 tasks; evaluation 2026-08-12. See source for repeat counts and wall-clock budget.	78.3% Reported settings Grok 4.6; source label: xhigh; xhigh effort; 23 tasks; evaluation 2026-08-12. See source for repeat counts and wall-clock budget.
ARC-AGI-3 · standardARC Prize · Standard Benchmark details ARC-AGI-3; Standard. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	2.11% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group xai-grok-4-6; variant xhigh; public-demo task scores are excluded
FrontierCode 1.1 · extended-grok-buildCognition · grok-build Benchmark details FrontierCode 1.1; grok-build. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-09-12.	55.02% Reported settings FrontierCode 1.1; extended; 150 tasks; grok-build; source effort low; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	59.62% Reported settings FrontierCode 1.1; extended; 150 tasks; grok-build; source effort medium; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.004	61.31% Reported settings FrontierCode 1.1; extended; 150 tasks; grok-build; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0047	Not reported
FrontierCode 1.1 · main-grok-buildCognition · grok-build Benchmark details FrontierCode 1.1; grok-build. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-09-12.	40.11% Reported settings FrontierCode 1.1; main; 100 tasks; grok-build; source effort low; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	46.06% Reported settings FrontierCode 1.1; main; 100 tasks; grok-build; source effort medium; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.006	48.01% Reported settings FrontierCode 1.1; main; 100 tasks; grok-build; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.007	Not reported
Terminal-Bench 4.0 · grokbuildTerminal-Bench · Grok Build Benchmark details Terminal-Bench 4.0; Grok Build. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	20.3% Reported settings Terminal-Bench 4.0.0; 66 tasks; Grok Build; effort high; reported margin ±3.1 percentage points, confidence level unspecified; release date 2026-08-12; board checked 2026-09-07	Not reported
Epoch · Chess Puzzles · v1.1.6Epoch AI · Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	40% Reported settings Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match; model=grok-4.6_high; run=NGQ4Q8Wrb4aroMu5RecTS6; mean accuracy; https://epoch.ai/data/benchmarks.csv	31% Reported settings Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match; model=grok-4.6_xhigh; run=RiyrdXiqrXT98mKM9nMsmu; mean accuracy; https://epoch.ai/data/benchmarks.csv
Epoch · FrontierMath Tier 4 · v2.0.0Epoch AI · Epoch AI Inspect; task 2.0.0; scorer verification_code Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. FrontierMath was developed with OpenAI funding and unequal access to part of the question set; this result is from Epoch's evaluation. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	31.71% Reported settings Epoch AI Inspect; task 2.0.0; scorer verification_code; model=grok-4.6_xhigh; run=bg3JRdd33iWfvRoGSvyXXE; mean accuracy; https://epoch.ai/data/benchmarks.csv
Epoch · FrontierMath Tiers 1–3 · v2.0.0Epoch AI · Epoch AI Inspect; task 2.0.0; scorer verification_code Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. FrontierMath was developed with OpenAI funding and unequal access to part of the question set; this result is from Epoch's evaluation. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	65.96% Reported settings Epoch AI Inspect; task 2.0.0; scorer verification_code; model=grok-4.6_xhigh; run=Zb2AKwa942D7dBTTcebAba; mean accuracy; https://epoch.ai/data/benchmarks.csv
Epoch · GPQA diamond · v1.0.11Epoch AI · Epoch AI Inspect; task 1.0.11; scorer choice Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	94% Reported settings Epoch AI Inspect; task 1.0.11; scorer choice; model=grok-4.6_high; run=hnP6Z88x8bPEBRZ2PW6kps; mean accuracy; https://epoch.ai/data/benchmarks.csv	93.18% Reported settings Epoch AI Inspect; task 1.0.11; scorer choice; model=grok-4.6_xhigh; run=9YJf2nviwjM6WVhw8qeJta; mean accuracy; https://epoch.ai/data/benchmarks.csv
Epoch · Mystery Game Puzzles · v1.0.4 · Tool submissionEpoch AI · Epoch AI Inspect; task 1.0.4; scorer agent_submitted_next_move_scorer Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	Not reported	34% Reported settings Epoch AI Inspect; task 1.0.4; scorer agent_submitted_next_move_scorer; model=grok-4.6_xhigh; run=9cXtChGrowgGDwa9AcAecB; mean accuracy; https://epoch.ai/data/benchmarks.csv
Epoch · OTIS Mock AIME 2024-2025 · v1.1.6Epoch AI · Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	97.78% Reported settings Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match; model=grok-4.6_high; run=jtfdCURnwfZzak4BSiBKzx; mean accuracy; https://epoch.ai/data/benchmarks.csv	99.17% Reported settings Epoch AI Inspect; task 1.1.6; scorer model_extracted_exact_match; model=grok-4.6_xhigh; run=7crLvnfQc987dC6mr3iGbf; mean accuracy; https://epoch.ai/data/benchmarks.csv
Epoch · SimpleQA Verified · v1.2.0Epoch AI · Epoch AI Inspect; task 1.2.0; scorer simpleqa_scorer Benchmark details Epoch AI internal evaluation. Mean accuracy under the published setting; exact task version and scorer stay separate. Related tasks share a family budget. Data by Epoch AI, CC BY. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	49.3% Reported settings Epoch AI Inspect; task 1.2.0; scorer simpleqa_scorer; model=grok-4.6_high; run=TpHYtuPG6kYZFvMK7Z6Hqj; mean accuracy; https://epoch.ai/data/benchmarks.csv	48.9% Reported settings Epoch AI Inspect; task 1.2.0; scorer simpleqa_scorer; model=grok-4.6_xhigh; run=dKCWFaoGCWJTnD5CPWQGru; mean accuracy; https://epoch.ai/data/benchmarks.csv
Vals AI · Finance Agent v2Vals AI · Vals FAB v2 shared six-tool harness; two-hour task limit Benchmark details Overall weighted rubric accuracy, averaged over three runs on 450 tasks. Dealbreaker failures and timeouts score zero. Related categories and all-pass scores are not additional votes. Three-model judge panel; private test set. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	53.68% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported
Vals AI · Code Migration v1Vals AI · Vals Code Migration shared agent harness Benchmark details Hidden behavior-test pass rate, equally weighted across 40 repositories (30 CLI and 10 COBOL-to-Java). Anti-cheat failures score zero. Code-quality ratings and language splits are not additional votes. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	44.57% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported
Vals AI · Legal Research Bench v1Vals AI · Vals Legal Research shared five-tool harness; three-hour task limit Benchmark details All-pass accuracy on 208 private legal research tasks. Every expert rubric check must pass; GPT-5.4 judges responses. Partial-credit scores and practice-area splits are not additional votes. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	48.08% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported
Vals AI · Excel Modeling Benchmark v1Vals AI · Vals EMB shared workbook agent harness; 3.5-hour task limit Benchmark details Mean of seven equally weighted task categories. Template submissions use recalculated Excel cell accuracy; scratch submissions use rubric checks judged by GPT-5.4 Mini XHigh. Partial credit does not mean task completion. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	62.73% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported
Vals AI · MMLU-Pro v1Vals AI · Vals MMLU-Pro five-shot chain-of-thought; regex answer grading Benchmark details Mean accuracy across 14 subjects under Vals five-shot prompting. Subject splits share one family and are not additional votes. Vals labels this benchmark archived; it is historical corroboration, not ongoing frontier evaluation. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	89.4% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-01","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported
Vals AI · Terminal-Bench 2.1Vals AI · Vals Terminus 2 harness; pass@1 on 89 tasks Benchmark details Overall task pass rate under the shared Terminus 2 configuration. All task tests must pass. Difficulty splits share one family. Fable 5 and Opus 5 mixed-checkpoint fallback runs are excluded. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	78.28% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported
Vals AI · Vibe Code Bench v1.1Vals AI · Vals modified OpenHands harness; five hours or 1,000 turns per app Benchmark details Mean app test pass rate on the canonical v1.1 test set. Browser Use executes workflow tests; each test requires at least 90% of its substeps. Automated judging can introduce error. Alternate agents are excluded. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	76.24% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":"OpenHands"}	Not reported
Vals AI · Harvey Legal Agent Benchmark v1Vals AI · Vals Harvey held-out legal file-work protocol; internet disabled Benchmark details Mean task resolution across two judges (GPT-5.5 Medium and Sonnet 4.6), requiring every criterion to pass. Criteria pass rates and practice-area splits do not add votes. Shared Harvey family; Fable 5 fallback run excluded. Unit: percent. Source collected 2026-09-12.	Not reported	Not reported	15.83% Reported settings {"checkpoint":"grok-4.6","evaluatorModel":"grok/grok-4.6","sourceUpdatedOn":"2026-09-10","reasoning_effort":"high","compute_effort":null,"reasoning":null,"temperature":0.7,"top_p":0.95,"max_output_tokens":null,"verbosity":null,"provider":"SpaceXAI","harness":null}	Not reported

What these comparisons mean

Each source reports its own effort labels. Matching labels do not establish equal compute budgets or identical fallback behavior. Open Reported settings to inspect the exact model name, agent and task coverage. Multiple results at one level remain visible rather than selecting the best. Source-reported None or Non-reasoning labels are not treated as evidence that an API supports those settings.

These measurements use their own dated source collection. The leaderboard fits identified configurations jointly; it contains no pooled model ratings. Explicit thinking budgets are separate configurations. Read the methodology.

Download all effort benchmark results · 7,424 results across 240 models · Collected 2026-09-12

Grok 4.6 · Medium effort benchmarks

Grok 4.6

Capability score evidence

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context